A Distributed Approach for the Optimal Power Flow Problem Based on ADMM and Sequential Convex Approximations

The optimal power flow (OPF) problem, which playsa central role in operating electrical networks is considered. Theproblem is nonconvex and is in fact NP hard. Therefore, designingefficient algorithms of practical relevance is crucial, thoughtheir global optimality is not guaranteed. Existing semi-definiteprogramming relaxation based approaches are restricted to OPFproblems where zero duality holds. In this paper, an efficientnovel method to address the general nonconvex OPF problemis investigated. The proposed method is based on alternatingdirection method of multipliers combined with sequential convexapproximations. The global OPF problem is decomposed intosmaller problems associated to each bus of the network, thesolutions of which are coordinated via a light communicationprotocol. Therefore, the proposed method is highly scalable. Theconvergence properties of the proposed algorithm are mathematicallysubstantiated. Finally, the proposed algorithm is evaluatedon a number of test examples, where the convergence propertiesof the proposed algorithm are numerically substantiated and theperformance is compared with a global optimal method.QC 20151211